xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<siraben>
mjlbach: it's possible to query Nixpkgs programmatically
<aterius>
I've done so for individual packages before
<aterius>
But during ZHF people usually post the hydra build blockers, but I'm not sure if those also include the packages that are disabled by being marked broken
<LnL>
I'm not sure, platforms vs broken is not applied consistently at all so you'd have to base things on hydra evaluations and not just metadata
<LnL>
with the first few darwin releases I used to go through ZHF and compare regressions with the previous versions
<LnL>
but that's quite a lot of work, I've only glanced at the overall status of the last releases TBH
<siraben>
Also many packages that would work on darwin don't have it darwin in meta.platforms
<LnL>
yeah, that as well as trivial fixes is what I used to do for things that where known to work on in previous releases
<LnL>
there's usually lots of low hanging fruit there like adding a clang flag, etc.
<siraben>
What's the recommend way to fix calls to gcc? doing something like CC=cc in the flags?
<LnL>
yeah, or even CC=$(CC)
<LnL>
depends ab a bit ofcorse, sometimes it's handcoded and needs patching in the Makefile
__monty__ has joined #nix-darwin
<LnL>
FYI I'm happily guide beginners through stuff like this, the group of darwin maintainers is pretty small and not everything needs crazy stdenv stuff to fix
philr has quit [Ping timeout: 256 seconds]
kalbasit has joined #nix-darwin
kalbasit has quit [Ping timeout: 256 seconds]
<aterius>
LnL: Thanks :) I might get to work on this then. My idea is more to show what is either marked broken, fails on hydra, or is not listed in supported platforms to give a ranked list of which derivations are causing the most darwin incompatibility
<aterius>
It would be nice to go through and fix all of the trivial changes/low-hanging fruit
xcmw has joined #nix-darwin
<abathur>
a little oblique, but I've wished we had more granular indicators. I don't know exactly what I mean, but some examples are being able to tell the difference between something marked unix and reported unix at the source site that is tested and known to work on macOS, vs projects where they say ~"I guess this should work on macOS but no one's ever reported"
<abathur>
or being able to tell that something marked unix was previously working on macOS+Nix and has slipped and stopped working at some point
<aterius>
That would be convenient, I'm trying to work around the lack of that by assuming that optimistically every package could conceivably work on darwin, but there's some fairly obvious stuff (gnome, xorg, wayland) that doesn't
<aterius>
I've noticed some packages declare gnome3 as a dependency when it clearly shouldn't because they are only used/tested on linux :(
<__monty__>
I prefer restriction by default.
<__monty__>
It's easy enough to override on a case by case basis.
<__monty__>
And this way a darwin user has to step up and more or less claim maintainership of the darwin part of the expression.
<__monty__>
Imo that's a better situation than just having a ton of unmarked but broken expressions for darwin and not having any maintainers.
__monty__ has quit [Quit: leaving]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<aterius>
Oh I totally agree with that, it would be easier to go through and just list all of the "marked broken" packages for darwin
<aterius>
I agree It would be nice to collate a database somewhere of legitimately marked broken packages vs "not packaged for darwin"
<LnL>
that's the platforms vs broken I was referring to, ideally platforms would always just match upstream support
{^_^} has quit [Ping timeout: 260 seconds]
Mic92 has quit [Ping timeout: 260 seconds]
Mic92 has joined #nix-darwin
<LnL>
alltho, I don't think removing platforms for things that break is very common so that might work relatively well for the things that worked at some point
<{^_^}>
#107768 (by rb2k, 2 minutes ago, open): bash-completion: can't be unpacked/built on macOS because of filenames
<rb2k>
@abathur mind posting a comment with your OS version and filesystem?
<aterius>
I think reading off packages that are marked broken will be insufficient, a lot are not marked available for darwin that are really just broken :(